Hi, I am creating Gaussian Splats with data from an Nvidia DriveWorks platform, and wondering if it’s possible to convert from the ftheta camera distortion model, to one supported by Colmap (Camera Models — COLMAP 3.12.0.dev0 documentation) or RealityCapture (RealityCapture Help).
This way I could supply the distortion parameters as priors, instead of having them estimated during pose estimation, and explore how to affects speed/quality.
Thanks!